From: Krekoski Ross (rosskrekoski@gmail.com)
Date: Tue Nov 25 2008 - 18:08:04 MST
Also, while I agree with what you're saying overall, I wonder if we
need to take into some consideration the total number of instruction
cycles for both scenarios, especially given how an evolutionary
algorithm will scale with population size over time.
Ross
On Wed, Nov 26, 2008 at 7:45 AM, Krekoski Ross <rosskrekoski@gmail.com> wrote:
> *AHEM*
>
> http://www.sl4.org/archive/0805/18859.html
> http://www.sl4.org/archive/0805/18858.html
>
> Needless to say I agree with what you're saying.
>
> Ross
>
> On Wed, Nov 26, 2008 at 5:10 AM, Matt Mahoney <matmahoney@yahoo.com> wrote:
>> --- On Tue, 11/25/08, Peter de Blanc <peter@spaceandgames.com> wrote:
>>
>>> Matt, your paper ends in the conclusion, "it seems
>>> that even though RSI is possible in a mathematical sense, it
>>> is too slow to be of practical use, or to compete with
>>> evolutionary systems, at least for fixed goals." What
>>> is the justification for this statement?
>>
>> The rate of information gain for RSI is O(log t). The information gain for evolution is O(t), or 1 bit per population doubling and selection.
>>
>> -- Matt Mahoney, matmahoney@yahoo.com
>>
>>
>
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:03 MDT